Does anyone have a complete working example on how to use PrefectResult? Its simple: Scrape a website for a link, if the link is different than the prefect result, download it and save the new link as the result. Otherwise exit with "success: no work to be done". I don't want this to be a local result because we plan to move these flows off of a windows agent and into docker agent. Thanks in advance!

For the first question on how to use the result, did you try:

@task(result=PrefectResult())

? this should be it

Hi @Kevin Kho, I'm stuck with the same question and it did not work for me. To be more specific, after adding result=PrefectResult() I expect there should be some way to retrieve that data. I run service locally (i.e. on the same node) with docker workers. I've set up caching, it seems to work on local runs. That's what I do:

from prefect import Client

client = Client()
client_flow_output_id = client.create_flow_run(
    parameters=dict(people=['one', 'two', 'three']),
    flow_id=flow_id,
)

I am not sure you can fetch the results of

PrefectResult

specifically. Maybe you can try the

get_task_run_result

task and invoke it using

get_task_run_result.run(…)

What are you trying to do? Are you trying to access it for just one task? Because persisting stuff in the KV Store might be easier to work with

KV store isn't an option, it does not scale and limitation of 10KB is too tight for me. I am testing if prefect can work in my scenario (I expect that before running a batch of jobs in multiple cases I'll need to tune the job parameters, when done need to check that it works in docker created for the task, then scale up): • I'd like to be able to run things locally (same env, different process) and return result for visualization / debugging. I can retrieve .result here, seems to work. Results are mixtures of images and tabular data. • I need to run on local docker executor two-three tasks and confirm they produce the same result • Then I need to scale it up. I'd like to cache results for some time so e.g. if something went wrong, I could check cached outputs during the next two weeks. I thought that cached result would be a nice option, but it is unclear how one can take retrieve result from a spawned process. If I can do that for S3Result - I can use S3Result, but it's still unclear if I can given the task result without knowing beforehand which key the task used for storing

whoops i gave the wrong link for KV store lol. reading the response

I can't make it work. Can you provide some minimal example? Caching as I mentioned works I've been checking graphql interface, but it appears that cache_key is empty - so it's unclear to me how one can start say from the flow run and collect intermediate computations

"flow_runs": [
          {
            "parameters": {
              "people": [
                "one",
                "two",
                "three"
              ]
            },
            "flow_id": "e78727e4-ac66-416b-bca1-65427daf58a6",
            "task_runs": [
              {
                "task": {
                  "name": "people"
                },
                "state": "Success",
                "cache_key": null,
                "state_result": null,
                "name": null,
                "tenant_id": "a7371ee2-0e8a-4309-9375-a017f502bb31"
              },
              {

I don't see even see keys being saved in the database, so it seems database will not be able to connect those

I think you are conflating that

cache_key

and the result. The result is for checkpointing. Caching is more about holding state in the Prefect Cloud/Server to not re-run a task. You should be looking for the

result

under the task run.

get_task_run_result

uses the

FlowRunView

here from the id that you give, which then goes here to a method to query, which then creates a Client to get the details for you. And then that Client instantiation pulls your stored API Key and uses it.

Thank you so much, that had shed light for me! I've found that queries also retrieve serialized_state for flow_run, which in turns contains result location field. Still digging into how client knows which endpoint to connect to.

Nice! Client just pulls the API KEY stored by default but you can point it explicitly